Have you ever had this moment?
You open your dashboard and stare at the retention curve for a long time, trying to read an answer out of it: has this product actually clicked, or not?
You look and look. The curve is still the same curve. You're still not sure.
I want to argue the exact opposite of that moment: whether your product has clicked is not something you should be looking for in the numbers. By the time you can see PMF clearly in retention and conversion, you should have felt it six months earlier.
The numbers are always the last to arrive
Whether a product has clicked, the dashboard is the last to know.
The real signals show up much earlier — and none of them live in your analytics:
A customer is furious because your thing broke, and comes at you angry. Someone willing to get mad about a product actually depends on it. Nobody gets worked up over something they could take or leave.
Someone mentions you and recommends you, completely unprompted. Spontaneous word of mouth can't be bought. It means your thing has entered someone else's vocabulary.
You yourself can't go a day without it. You're the first user. You know better than anyone whether it actually solves the pain.
Once those three signals appear, PMF is already there. And it takes months for them to slowly surface in the numbers.
Waiting for the numbers is choosing, on purpose, to learn the answer six months late.
Why we'd rather wait for the late number
This is the part I really want to get at.
If feeling is earlier and more accurate, why do most people still stare fixated at the numbers?
Because looking at the numbers means you don't have to be responsible.
"I think this thing has clicked" — the moment you say it out loud, you own it. You made the call. If it's wrong, it's on you.
"The data shows this thing has clicked" — that sentence is so much more comfortable. You're just a narrator. You point at the curve; responsibility sits with the data, not with you.
Hiding inside the data is, at its core, quietly swapping "I think" for "the data thinks" — outsourcing your judgment to a spreadsheet so that if it's wrong, it's not your fault.
But at the earliest stage, you might have only a few dozen users. Any single number can be thrown off by one random fluctuation. The only thing that can save you then is exactly that judgment — "as the person who understands this pain best, do I feel it has clicked?"
Hand that judgment away in exchange for a report that looks objective, and you're switching off your own perception at the moment you need it most.
So when should you look at the numbers?
I'm not saying numbers are useless.
After you scale — thousands, tens of thousands of users — personal feel can't cover it all, and numbers become invaluable. They show you the whole picture that one person can't see.
But that's the "already clicked, now grow it" stage.
At the "has it clicked or not" stage, the numbers can't give you the answer — only you can. Use the wrong tool for the stage and you get in trouble: worship numbers too early and you'll dare to decide six months late, or get fooled by small-sample noise into the wrong direction; rely only on feel too late and you'll miss the real problems that live at scale.
The skill worth training isn't "numbers or no numbers." It's telling the difference: is this product right now in the stage that needs feel, or the stage that needs numbers?
Finally
In my own experience building things, every time a product truly took off, the order was the same: I felt it first, and the numbers caught up to confirm it later. Never the other way around.
Whether something still in its cradle is worth betting on — that's not a call the data can make for you.
It can only be made by the person closest to it, who understands the pain best.
That person is you. Don't hand that judgment to a spreadsheet that shows up six months late.
Top comments (0)